Subject: Re: Open letter to those who believe in a right to free software
From: Bernard Lang <Bernard.Lang@inria.fr>
Date: Sun, 7 Nov 1999 01:54:20 +0100


This may seem out of topic... but I do believe that any remark about
the creative process in mathematics is relevant for informatics (I
just attended a conference where I was told that computer science is a
bad terminology implying it is not a science).

Having a proper understanding of creativity processes seems essential
to understand what economic strcture may be favorable to technological
growth, and thus to economic growth.

Now what is good globally may not be the optimal solution for any
given individual or business.

On Wed, Nov 03, 1999 at 11:58:43AM -0500, Ben_Tilly@trepp.com wrote:
> You mean what lessons should we draw from it?  The biggest
> ones have to do with the importance of keeping lines of
> communication open between different groups if progress is
> going to happen.  The importance of this improves if your
> endeavor (like with free software) is using a distributed
> development model.
> 
> Don't just work out how you will do things.  Keep track of how
> other people are doing then and know how to explain your
> stuff to them.

I am probably feeble-minded ... but as a computer scentist I find this
beyond my abilities in CS alone. And I did often sit in talks I did
not understand (hence I usually take some work to read, just in case).
I do know colleagues that can sit trough a greater variety of talk
than I can myself, but none that can listen to anything, especially if
sufficiently technical, which is unavoidable when working at the
forefront of new research.

  But I do believe that somehow I did manage some contributions, and
so do other people who share my failings.

   Keeping lines open is not easy, and not cheap. I had the experience
of switching to a new CS area. The first thing I had to do was to
learn their language, their concepts, so that I would understand them
and they would me. It took more than a year, just to begin to be able
to conceptualize their way.

  So I do not believe that math is unique. And I do not believe that
the incriminated behavior prevents producivity, as long as some people
make the effort of transfering the knowledge (keeping the lines open),
sometimes at the expense of being original contributors.

> My wife's field reports on biology, as well as experiences in
> other sciences indicate that this situation is specific to math.

I strongly doubt that, except possibly for extremely talented people,
and even then. When you read the proceedings of some large spectrum
conferences, it is very unlikely that people uderstand even half of
what is presented (and I am not talking about CS ... I occasionally
look at other stuff).

> >  - that math is indeed very difficult ?
> 
> Anything can be difficult if you make it so.

sure ... but it does not imply that anything can be made easy.
Or, at least it may take a very long time to do that.
 
> Maintaining lines of communication on technical subjects
> requires work both on the part of would-be listeners and on the
> part of the speaker.  This effort is worthwhile and failure by
> either side to recognize this fact is dangerous to the long-term
> health of your endeavor.

It is just not possible when at the forefront of research. Hard enough
to find a common ground with the specialists of your narrow field.  I
have also the experience that trying to induce people to get a
different view of what they already know is very hard and may take a
long time (unless you are a recognized guru, whose word is not to be
doubted).

> Plus nobody knows how much is being lost.

that is true, for any system ... and therefore irrelevant

> The question of math being lost is not trivial.  For instance I
> once figured out a nice way to generate answers to such
> questions as, "Which polynomial with integer coefficients
> has square-root(2) plus cube-root(3) as a root?"

I am not sure what is the general problem you have in mind, but for
this specific example it took me 5 mn of high-school algebra to find
an answer: x^4-10x^2+1 (only 4'th degree polynomial)
  What is the general statement of the problem ?
[sorry if that is really irrelevant to this list]

> This holds both in the study of symmetric polynomials and
> algebraic number theory, both originally motivated by the
> solution to my problem, both having forgotten the solution
> to my problem, and both having forgotten that they ever
> had any connection to each other!

Forgetting is part of learning. You weed out the unessential ... with
a risk, so as to keep things manageable. That is what waste paper
baskets are for. I should know, I am very bad at weeding out, I tend
to consider everything as potentially useful, which it is, and my
office is a mess.

> >  - ... and it might be woth investigating how and why ?
> >  - that most mathematicians are not really contributing,
> >    only some of the best ones ?
> 
> This is unfortunately true.  The lesson to be drawn is the
> importance of keeping lines of communication open
> and not getting isolated.  But it takes effort.

yes... some individuals are specialists of that (it is a collective
endeavor, not necessarily the responsibility of each single person).
And it is often poorly recognized by the community (not original).

> >  - ... and that you know how to identify the good ones (or you don't) ?
> 
> You don't.

Yes ... you don't ... and that is why you have to cope with too large
a community. Like a gold digger who would prefer to dig only where
gold is to be found ...

> >  - ... and that the other are completely useless ?
> >    or possibly they fulfill another role ?
> >
> They fill another role.  They teach calculus.

... yes
    but possibly they explore paths that are not successful, or they
do community service, like organizing conferences, or they play the
role of sparring partners for the better ones who need to try out
ideas. ... or they maintain communication lines. The world is not
simply divided between competent and incompetents. There are many
roles.

> Anyways I don't think that mathematics is a very good
> model for free software to follow.

I do not idealize mathematics, nor any community. I know the
mathematicians of his time let Abel starve to death... All communities
are human. But I still believe it does work extremely well when
considering the actual achievments, the large picture. And I wonder
why you seem to hold this community in such low esteem.

  This being said, I am not advocating that it should be forbidden to
do commercial software, or free software in a commercial environment
(actually there is good math being done in commercial environment
too).  I am just saying that the academic model does give good results
in the long run. And that there are many technical reasons to believe
that the development of software has very similar requirements.

  A nd since you insist on communication, which I agree with ... I
think it is a lot easier in the academic world than in the commercial
world ... though academia appears more and more perverted by
proprietary thinking.

> Yet I can point at major results
> that are widely distrusted within mathematics!  I am not
> just talking about mistakes in random papers.  I am
> talking about thinks like the classicification of finite
> groups (an effort that took decades and is being
> redone because the people who did it can't read and
> don't believe their own proof) and the proof of the four
> color theorem (the debate about which literally comes
> down to deciding whether the bugs in the program they
> wrote for their analysis came up when they ran it).

so what ... redoing programs several times is considered the right
thing to do by many people. You are supporting my views.
 
> I hold free software development to higher standards
> than that,

than what ... nothing wrong with the above. You are only saying that
these problems are hard, and they did not get them completely right on
the first try, and that the code - I mean the proof, but Curry-Howard
say it is the same - was a bit messy. But that is only natural.

Amicalement

  Bernard


-- 
Bernard.Lang@inria.fr             ,_  /\o    \o/    Tel  +33 1 3963 5644
http://pauillac.inria.fr/~lang/  ^^^^^^^^^^^^^^^^^  Fax  +33 1 3963 5469
            INRIA / B.P. 105 / 78153 Le Chesnay CEDEX / France
         Je n'exprime que mon opinion - I express only my opinion
                 CAGED BEHIND WINDOWS or FREE WITH LINUX